Summary

A major North American media and entertainment company partnered with Marlabs to modernize its legacy ad campaign planning platform. The existing optimization engine relied on a commercial solver that generated campaign plans with nearly zero acceptance rates and required extensive manual input. By implementing a machine learning–driven optimization system built on historical campaign data, Marlabs enabled smarter planning, improved KPI alignment, reduced operational friction, and eliminated costly licensing fees.

The new architecture introduced predictive analytics and a multi-objective optimization engine that directly targeted business outcomes such as impressions, reach, and gross rating points (GRP). As a result, the organization gained a scalable, data-driven campaign planning capability that increased inventory utilization, strengthened advertiser relationships, and delivered significant cost savings.

Challenge

Replacing a costly, ineffective ad campaign optimization system

A major media organization relied on a legacy campaign planning tool that consistently produced low-quality advertising plans and required expensive licensing. The system’s reliance on a commercial solver and heavy manual input created inefficiencies that prevented the organization from optimizing toward meaningful advertising KPIs such as impressions, GRP, and reach.

Because the platform lacked learning capabilities and relied heavily on manual inputs from sales teams, campaign planning often resulted in inaccurate recommendations and poor acceptance rates. This created operational friction, reduced advertiser confidence, and limited the company’s ability to maximize revenue from its advertising inventory.

Solution

Delivering intelligent ad campaign optimization through ML and open-source tools

Marlabs designed and implemented a machine learning–driven campaign optimization platform that uses historical campaign performance data to generate predictive insights for future ad planning. The solution leveraged a cloud-based architecture and replaced the legacy solver with a multi-objective optimization engine capable of aligning campaign parameters directly with business KPIs.

The new platform minimized reliance on manual inputs by predicting optimal parameters for campaign planning and automatically incorporating them into the optimization process. Through a combination of predictive analytics, advanced optimization techniques, and improved workflow design, the system enabled the client to generate higher-quality campaign plans with greater efficiency and accuracy.

Historical Data Analysis & Feature Engineering

Marlabs began by analyzing historical advertising campaign data to identify the variables most closely correlated with successful campaign outcomes. This phase focused on extracting meaningful insights from past performance data to inform predictive modeling and optimization strategies. By evaluating patterns across impressions, reach, GRP performance, and campaign acceptance rates, the team established a robust data foundation for the new planning system.

Through feature engineering, the team transformed raw campaign data into structured inputs suitable for machine learning models. These engineered features captured relationships between campaign parameters and performance metrics, enabling the system to predict optimal planning variables with greater precision. This data preparation step ensured that the subsequent models were built on high-quality, relevant signals rather than incomplete or noisy inputs.

Machine Learning Model Development

In the second phase, Marlabs developed machine learning models capable of predicting optimal parameter estimates for new advertising campaigns. These models were trained on historical campaign data to identify patterns that drive successful advertising outcomes.

The models were evaluated and refined through iterative testing to ensure their predictions aligned with the client’s business KPIs. By learning from past campaign performance, the system could recommend parameter configurations that improved campaign acceptance rates and better aligned advertising plans with measurable outcomes such as impressions and GRP delivery.

Optimization Engine Replacement

With predictive insights established, the team replaced the legacy commercial solver with a modern, open-source multi-objective optimization engine. This new system was designed to integrate machine learning outputs directly into campaign planning, enabling automated optimization across multiple performance metrics simultaneously.

The optimization engine incorporated iterative processing and human-in-the-loop checkpoints to maintain operational oversight while improving automation. By aligning optimization objectives with key business KPIs, the platform generated higher-quality campaign plans while eliminating reliance on expensive proprietary software.

User Input Minimization & System Integration

In the final phase, Marlabs streamlined the campaign planning workflow by minimizing the amount of manual input required from users. Machine learning predictions were used to auto-populate key planning parameters, significantly reducing the complexity of the user interface and improving usability.

The simplified workflow reduced the likelihood of human error and accelerated campaign setup times for sales teams. By integrating predictive modeling, optimization algorithms, and user-friendly interfaces into a unified system, the platform delivered a modernized planning environment that empowered teams to generate more effective campaign strategies with less effort.

Results

The ML-driven campaign optimization platform delivered substantial improvements in planning accuracy, operational efficiency, and cost performance. By leveraging predictive analytics and automated optimization, the organization was able to generate campaign plans that aligned more closely with advertising KPIs and business outcomes.

The system reduced reliance on manual inputs and introduced data-driven parameter estimation, improving both internal confidence in campaign recommendations and advertiser satisfaction with campaign outcomes. In addition, the shift to an open-source architecture eliminated costly licensing fees associated with the previous commercial solver.

Impact

• Improved GRP Delivery: Optimized ad placements increased inventory utilization and improved overall campaign performance.
• Higher Revenue Potential: Even small efficiency gains translated into millions of dollars in additional annual revenue.
• Cost Savings: Replacing the commercial solver eliminated more than $250,000 per year in licensing costs.
• Better User Experience: Reduced input requirements simplified campaign planning and decreased the likelihood of human error.
• More Accurate Campaign Plans: Predictive parameter estimation improved plan acceptance rates and planning accuracy.
• Stronger Advertiser Relationships: Increased impressions and improved CPM performance enhanced advertiser satisfaction and demand.